[tex-live] texdoc error

Manuel Pégourié-Gonnard mpg at elzevir.fr
Wed Sep 3 00:28:05 CEST 2008


Reinhard Kotucha scripsit (03.09.2008 00:09)
> the child process.  If this is not clearly specified, it's at least
> not unlikely that different versions of Windows behave differently.
> 
And even if there's something clearly specified... I heard funny things
recently about how MicroSoft "implements" the ntfs spec, though they wrote it
themselves...

> But it's good to know that the wrapper is not the culprit, though.
> 
Hum, actually, it sort of is. I don't mean the wrappers should be written
another way, and I don't see any solution, but the fact remains that if you run

texlua c:\path\to\texdoc.tlu texlive-en

it works, but not (with AR9 on certain windows installations) if you run

texlua c:\path\to\tl-w32-wrapper-script.tlu texdoc texlive-en

It all boils down to the following:

% wrapper
kpse.set_program_name("texlua")
os.spawn {"texlua", "c:\\tmp\\program.tlu"}

% program
os.execute {"acrord32", "c:\\tmp\\pdf-file.pdf"}

(with acrord32 supposed to be in the path and to be AR9, and pfd-file.pdf an
existing pdf file). I run

texlua wrapper.tlu

and AR crashes badly (internal error). If I either:
- run another viewer in place of acrord32, or
- remove the first line of wrapper.tlu, or
- change the seconde line to use os.execute

it works.

I forwarded this example to the luatex maintainers, but Hans cannot reproduce
the problem with his machine. I'll try installing luatex 0.29 on my windows to
see if the luatex version explains that Hans has no problem while everyone
here does.

Since it looks like you know the story of os.spawn better than I do, if you
have any idea, be welcome.

Manuel.


> Regards,
>   Reinhard
> 


More information about the tex-live mailing list